Rapid transit - getting data to
  public health agencies

May 2005
Feature Story

Most laboratory directors focus on producing results for the ordering physician, to be used in caring for the patient immediately. This is indeed our most important job. However, many clinical labs today are overlooking other important uses of those data.

Information on the infections identified in laboratories—by culture, serology, or other means—is of critical importance to all of us in controlling the spread of communicable disease. Reporting that information promptly to the local public health agency not only is a good idea but also is required by law. We now have vastly improved mechanisms to make it happen—electronic linkages between our laboratory information systems and public health information systems.

Why this is a plus for the lab:

• Results are sent to public health as soon as they’re available rather than days later. Public health staff can begin evaluating cases in the laboratory’s local community much more promptly. Some organisms must be reported to public health within 24 hours; this is much more likely to happen when the routing is automatic and continuous.
• Every case that meets defined criteria is sent automatically.
• Laboratory management is better assured that it is in compliance with the reporting law. In addition, some public health jurisdictions, such as New York state, require that laboratories report electronically. Others, such as the California lead program, have required electronic reporting of particular analytes.
• Though the HIPAA law exempts public health reporting from the patient-permission requirement, records of what has been sent must be kept. An electronic interface does this automatically—and more reliably and completely than a manual reporting system.
• Manually reporting legally reportable laboratory data consumes considerable staff time. A study at the Washington State Department of Health found that it costs laboratories between $.50 (for very high-volume laboratories) and $5 (for a more typical lab) in staff time to send a lab report manually to the public health agency.

What types of information should you send? First, reportable diseases—every case of Salmonella, Chlamydia, viral hepatitis, and so on. In California, more than 80 infectious diseases are legally reportable. As soon as public health personnel receive your report of a hepatitis case, they will call to ask for the liver enzymes, so I advise sending them with the original report.

Second, send all antimicrobial susceptibility results (for all organisms, not just reportables). In the past, public health agencies collected summarized antibiograms from the hospitals in their jurisdiction. However, different institutions use different rules to create those—with different denominators. By obtaining raw data continually from all local laboratories, the public health agency will be able to calculate resistance patterns consistently and comprehensively, and to look for subtle trends that may be affecting one region of the community or one type of institution. Patient identifiers are not required for this analysis, but a unique hashed patient code is helpful (to recognize an organism that has been cultured two or three times in a single patient and to avoid counting that multiple times in the statistics).

Third, send a de-identified order transaction for every blood culture, CSF cell count, and other tests indicative of a workup for infectious disease. The pattern of certain test orders may be a useful indicator of disease outbreaks. For example, one index of influenza season in our community is that the blood culture machines at some of our busier hospitals fill up. Rather than require the laboratory information system to maintain a daily count of blood cultures, the simplest operational procedure may be to send a de-identified order transaction for every culture. Likewise, cerebrospinal fluid cell count orders (and perhaps results) and stool culture orders may be useful.

Many health departments now provide Web pages for laboratories to type in reportable disease cases. Though better than paper reports—12 laboratories in Los Angeles County now report this way—these Web pages do not provide most of the advantages of full electronic reporting between the lab information system and public health information system. Unfortunately, every laboratory has devised its own codes to define tests resulted, organisms, and specimen sources. Before the connection with public health is activated, a translation table must be built from these unique, laboratory-specific codes to industry standard codes (LOINC for result names and SNOMED for organism names and specimen sources).

A number of steps are then required to transmit a laboratory result, or order, from the LIS to the public health system:

• The sending program examines all orders and results.
• In some implementations, all of these are forwarded to a filtering system. In others, software on the LIS does the filtering (selecting out those results and orders that must be sent to public health).
• Data are formatted into an HL7 standard message. The most generally accepted version is 2.3.1. The implementation guide is at www.cdc.gov/phin/messaging/guides/ELR231ImpGuide042103.pdf. In some cases, when the LIS is incapable of producing HL7, a less satisfactory flat-file format is used.
• Data are sent via a secure link to public health or to an intermediary. I recommend a link allowing for a continuous data stream (such as a virtual private network) rather than intermittent transmissions (such as secure file transfer protocol).
• At public health (or in some cases in the sending laboratory information system or at the intermediary), lab-specific test codes, result codes, and specimen source codes are translated into LOINC and SNOMED.

Thus, a number of configurations for systems lie between the LIS and the public health information system. The simplest is a security channel—all processing is done on the LIS or on the public health information system. Next is a filtering system—all results from the LIS flow to the filter, which sorts out the cases to be sent on, often performs LOINC and other translations, and sends the appropriate results to public health. Two examples of such devices are the Health System Resident Component provided by the Real-time Outbreak Detection System laboratory of the University of Pittsburgh, and the Public Health Information Link processor developed by Atlas Development. A third possibility is a data collator/intermediary. All results flow to the intermediary, which processes information through a variety of algorithms, then sends appropriate cases to public health. This is the strategy of the Cerner Health Sentry offering.

While this discussion has focused on clinical laboratory—in particular microbiology—findings, similar principles apply to other types of electronic laboratory reporting, such as surgical pathology data sent to tumor registries and serology results sent to immunization registries.

To my knowledge, the first electronic laboratory reporting interface to public health (from a surgical pathology system to a tumor registry) became operational in 1977. Though progress since then has been gradual, we are now seeing an acceleration of connections. New York State has 65 laboratories connected (40 via HL7 and 25 via flat file). Several other jurisdictions are making progress.

Contact your own public health agency to explore how you can connect. Is your local health department, state agency, or an intermediary facilitating connections? (For example, in Indiana the Regenstrief Institute is facilitating connectivity.) What types of data do they want—reportables only or also susceptibilities and syndromic indicators? Do they have funding to help cover your out-of-pocket costs—for example, vendor license fees?

Above all, make your contribution to protecting the health of your community.

Dr. Aller is director of bioterrorism preparedness and response for Los Angeles County Public Health Acute Communicable Diseases.